HIG Doctor packages Apple's Human Interface Guidelines into agent-readable retrieval and audit tooling.
The project combines a structured HIG corpus, an MCP server for retrieval, and a compliance scanner that produces severity-bucketed findings across modern UI frameworks.
System
- Apple HIG topic corpus exposed through skills, raw Markdown, and llms.txt
- Rule definitions grouped by framework, severity, and UI concern
- Retrieval surface for coding agents that need design-system context
- Scanner output designed for review, CI, and agent repair loops
Runtime
- stdio MCP server for agent retrieval and audits
- CLI/browser audit flow with JSON, stdout, and export modes
- GitHub Action wrapper for CI use
- Annual refresh workflow for WWDC and guideline changes
Proof
- Multi-framework detection rules cover SwiftUI, UIKit, React, CSS, Android, Flutter, and others
- Context-aware matching reduces false positives in print, motion, and focus-visible cases
- Severity gates allow CI to block only critical or serious violations
- Comment stripping preserves string literals while avoiding dead-code findings